rio relational indexing for object recognition n.
Skip this Video
Download Presentation
RIO: Relational Indexing for Object Recognition

Loading in 2 Seconds...

play fullscreen
1 / 18

RIO: Relational Indexing for Object Recognition - PowerPoint PPT Presentation

  • Uploaded on

RIO: Relational Indexing for Object Recognition. RIO worked with industrial parts that could have - planar surfaces - cylindrical surfaces - threads. Review of Alignment. Alignment is the most common paradigm for matching 3D models to either 2D or 3D data. The steps are:.

I am the owner, or an agent authorized to act on behalf of the owner, of the copyrighted work described.
Download Presentation

PowerPoint Slideshow about 'RIO: Relational Indexing for Object Recognition' - kristen-wong

Download Now An Image/Link below is provided (as is) to download presentation

Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author.While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server.

- - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - -
Presentation Transcript
rio relational indexing for object recognition
RIO: Relational Indexing for Object Recognition
  • RIO worked with industrial parts that could have
  • - planar surfaces
  • - cylindrical surfaces
  • - threads
review of alignment
Review of Alignment

Alignment is the most common paradigm for matching

3D models to either 2D or 3D data. The steps are:

1. hypothesize a correspondence between a set of model

points and a set of data points

2. From the correspondence compute a transformation

from model to data

3. Apply the transformation to the model features to

produce transformed features

4. Compare the transformed model features to the image

features to verify or disprove the hypothesis

2d 3d alignment
2D-3D Alignment
  • single 2D images of the objects
  • 3D object models
  • - full 3D models, such as GC or SEV
  • - view class models representing characteristic
  • views of the objects
view classes and viewing sphere
View Classes and Viewing Sphere
  • The space of view points can be
  • partitioned into a finite set of
  • characteristic views.
  • Each view class represents a set of
  • view points that have something
  • in common, such as:
  • 1. same surfaces are visible
  • 2. same line segments are visible
  • 3. relational distance between pairs of them is small



3 view classes of a cube
3 View Classes of a Cube

1 surface 2 surfaces 3 surfaces

object representation in rio
Object Representation in RIO
  • 3D objects are represented by a 3D mesh and set of 2D view classes.
  • Each view class is represented by an attributed graph whose
  • nodes are features and whose attributed edges are relationships.
  • For purposes of indexing, attributed graphs are stored as
  • sets of 2-graphs, graphs with 2 nodes and 2 relationships.

share an arc

coaxial arc



rio features
RIO Features

ellipses coaxials coaxials-multi

parallel lines junctions triples

close and far L V Y Z U

rio relationships
RIO Relationships
  • share one arc
  • share one line
  • share two lines
  • coaxial
  • close at extremal points
  • bounding box encloses / enclosed by
hexnut object
Hexnut Object

What other features

and relationships

can you find?

graph and 2 graph representations
Graph and 2-Graph Representations

1 coaxials-



1 1 2 3

2 3 3 2


2 ellipse

e e e c


3 parallel



relational indexing for recognition
Relational Indexing for Recognition

Preprocessing (off-line) Phase

  • for each model view Mi in the database
  • encode each 2-graph of Mi to produce an index
  • store Mi and associated information in the indexed
  • bin of a hash table H
matching on line phase
Matching (on-line) phase
  • Construct a relational (2-graph) description D for the scene
  • For each 2-graph G of D
  • Select the Mis with high votes as possible hypotheses
  • Verify or disprove via alignment, using the 3D meshes
  • encode it, producing an index to access the hash table H
  • cast a vote for each Mi in the associated bin
  • The matched features of the hypothesized object are
  • used to determine its pose. Pose is computed from
  • correspondences between 2D and 3D points, lines,
  • and circles.
  • 2. The 3D mesh of the object is used to project all its
  • features onto the image using perspective projection
  • and hidden feature removal.
  • 3. A verification procedurechecks how well the object
  • features line up with edges on the image, using a
  • Hausdorf distance between expected and existing edges.
rio verifications
RIO Verifications